Palynological evidence for late Miocene stepwise aridification on the northeastern Tibetan Plateau

Holding a climatically and geologically key position both regionally and globally, the northeastern Tibetan Plateau provides a natural laboratory for illustrating the interactions between tectonic activity and the evolution of the Asian interior aridification. Determining when and how the late Miocene climate evolved on the northeastern Tibetan Plateau may help us better understand the relationships among tectonic uplift, global cooling and ecosystem evolution. Previous paleoenvironmental research has focused on the western Longzhong Basin. Late Miocene aridification data derived from pollen now require corroborative evidence from the eastern Longzhong Basin. Here, we present a late Miocene pollen record from the Tianshui Basin in the eastern Longzhong Basin. Our results show that a general trend toward dry climate was superimposed by stepwise aridification: a temperate forest with a rather humid climate existed in the basin between 11.4 and 10.1 Ma, followed by a temperate open forest environment with a less humid climate between 10.1 and 7.4 Ma, then giving way to an open temperate forest–steppe environment with a relatively arid climate between 7.4 and 6.4 Ma. The vegetation succession demonstrates that the aridification of the Asian interior occurred after ∼ 7–8 Ma, which is confirmed by other evidence from Asia. Furthermore, the aridification trend on the northeastern Tibetan Plateau parallels the global cooling of the late Miocene; the stepwise vegetation succession is consistent with the major uplift of the northeastern Tibetan Plateau during this time. These integrated environmental proxies indicate that the long-term global cooling and the Tibetan Plateau uplift caused the late Miocene aridification of the Asian interior.


Introduction
As the latter stage of the global Cenozoic cooling, the Neogene was a critical period for northern hemispheric aridification, especially for the marked aridification of the Asian interior. Establishing when, and how, this process of aridification began and evolved is therefore vital for elucidating the interactions among tectonic uplift, global cooling and ecosystem evolution. Although there is compelling evidence for the aridification of the Asian interior, there is no consensus concerning its evolution and driving mechanisms. For instance, previous researchers have suggested that the aridification of the Asian interior began in the late Miocene, based particularly on biological and isotopic evidence (Andersson and Werdelin, 2005;Cerling et al., 1997;Dettman et al., 2001;Eronen et al., 2012;Quade et al., 1989;Wang and Deng, 2005;Zhang et al., 2012). However, others have argued that the process of Asian interior aridification may have begun in the early Miocene (22 Ma) or even earlier (in the late Oligocene), as inferred from Miocene or Oligocene eolian deposition (Guo et al., 2002(Guo et al., , 2008Qiang et al., 2011;Sun et al., 2010). The particular driving mechanisms of such aridification also remain enigmatic. Up until now, the tectonic uplift of the Tibetan Plateau (TP), global cooling and land-sea distributions have been suggested as the major drivers Gupta et al., 2004;Kutzbach et al., 1993;Liu and Yin, 2002;Miao et al., 2012;Molnar et al., 2010) H a iy u a n F a u lt H a iy u a n F a u lt L iu p a n s h a n F a u lt L iu p a n S h a n F a u lt there is little consensus about which one is the most important driver. We focused on the region of the northeastern TP to explore the nature of the interactions between tectonics and climate. The geographically extensive Longzhong Basin, consisting of a series of subbasins, is located in the northeastern TP. These subbasins present a continuous record of mammalian fossil-rich Cenozoic sediments, recording the effect of TP uplift on regional climates (Fang et al., , 2005GRGST, 1984;Li et al., 2006Li et al., , 2014, as well as the effect of the global cooling. On the other hand, it lies in the so-called monsoonal triangle, a transition zone from a warm-humid Asian monsoonal climate to a dry-cold inland climate and to the alpine climate of the TP (Li et al., 1988(Li et al., , 2014 (Fig. 1a). Its particular geological and geographical characteristics make it sensitive to document the aridification history of northern China. As a field laboratory for studying tectonic-climate interactions (Molnar et al., 2010;Tapponnier et al., 2001), the Longzhong Basin might be the most promising for distinguishing TP uplift and associated environmental change.
As a reliable paleoenvironmental proxy, pollen has been used to reconstruct past climates because of its abundance and excellent preservation within sediments. Previous research has demonstrated that the Tianshui Basin, as a subbasin of the Longzhong Basin, exhibits a typical late Miocene lacustrine-fluvial sedimentary succession containing abundant pollen . Here we reconstruct a high-resolution palynological record from the well-dated Yaodian Section, located in the southern part of the Tianshui Basin. Our results not only provide new evidence for the evolution of vegetation in the late Miocene and climate change on northeastern margin of the TP but also shed new light on the aridification of the Asian interior.

Geological and geographical settings
The rhomboid-shaped Longzhong Basin, which is one of the largest intermountain and fault-controlled sedimentary basins on the northeastern TP, is geographically delineated by the left-lateral strike-slip Haiyuan Fault to the north, the Liupan Shan Fault to the east and northeast, the Laji Shan Fault to the southwest, and the Western Qinling Fault to the south (Fig. 1b). The Tianshui Basin, one of its subbasins, is located in the southeastern part of the Longzhong Basin (Fig. 1b). It has witnessed the continuous deposition of mammalian fossil-rich Cenozoic sediments from the surrounding mountains; these sediments record the interactions between mountain uplift, erosion and climate change (Alonso-Zarza et al., 2009;Li et al., 2006;Liu et al., 2015;Peng et al., 2012Peng et al., , 2016. At present, the East Asian monsoon influences this region, engendering a semi-humid, warm-temperate, continental monsoon climate, characterized by relatively hot, humid summers and cold, dry winters. The mean annual temperature and mean annual precipitation of this area are ∼ 11 • C and 492 mm, respectively, with rainfall concentrated mainly in summer and autumn (Fig. 1c). The modern nat- ural vegetation in this region is warm-temperature forestgrassland. Warm grasslands are distributed in the valleys, and consist mainly of Arundinella hirta, Spodiopogon sibiricus and Themeda triandran. Shrubs such as Zizyphus jujuba, Sophora viciifolia and Ostryopsis davidiana are found on the hillsides. Trees, including Quercus liaotungensis, Pinus tabulaeformis, P. armandi and Platycladus orientalis, grow in the mountains (Huang, 1997).
The selected Yaodian Section (105 • 55 E, 34 • 38 N) is located in the southern part of the Tianshui Basin (Fig. 1d). The Neogene sequence in section is capped by loess and lies unconformably on top of the Paleogene Guyuan Group. It has been divided into the Ganquan Formation (Fm), the Yaodian Fm and the Yangjizhai Fm, in sequence upwards . In this study, our research mainly focuses on the late Miocene Yaodian Fm and Yangjizhai Fm. Based on a determination of lithology and sedimentology, the Yaodian Fm can be divided into three principal strata. The lower stratum consists of massive fine gravel sandstone, sandstone and brown silty mudstone, occasionally with thin brown mudstone or interbedded paleosols, which can be considered fluvial channel deposits (Fig. 2e). Abundant teeth of Hipparion weihoense, Cervavitus novorossiae, Ictitherium sp. and their bone fragments were excavated from this stratum. The middle stratum of the Yaodian Fm consists of the interbedding of siltstone or fine sandstone with mudstone intercalated with paleosols, overlying the fluvial channel deposits. The assemblage's characteristics are typical of floodplain depo-sition (Fig. 2d). The upper stratum of the Yaodian Fm is characterized by rhythmic cycles composed of grey or brown mudstone or sandy marlite and intraclastic marl intercalated with brown siltstone and mudstone and contains fossil algae and gastropods; this section is representative of shallow lake deposition ( Fig. 2a and c). The upper stratum is common throughout the basin, and is analogous to the "Zebra Bed" stratum found in the Linxia Basin in the western Longzhong Basin (Li et al., 1995). The Yangjizhai Fm is principally composed of reddish-brown mudstone or silty mudstone and yellowish brown calcrete or calcareous mudstone, with scattered sandstone or grey mudstone and marlite. These sediments were deposited under strong evaporative conditions in distal floodplain to palustrine environments (Fig. 2b). Previous paleomagnetic investigations have indicated that the Yaodian Fm ranges from 11.67 to 7.43 Ma in age, and that the Yangjizhai Fm dates from 7.43 to 6.40 Ma, with both of these ranges being consistent with the formations' biostratigraphic ages .

Materials and methods
Most of the samples came from lacustrine mud deposits and fine grain size intercalations found in floodplain and fluvial channel deposits. Because the lower 10 m of the Yaodian Fm consists of coarse gravel sandstone, and it was difficult to find fine-grained sediments therein, this part of the formation was not sampled. A total of 200 samples were processed for palynological analysis. For each sample, > 100 g of sediment was washed in 20 % HCl, soaked in 39 % HF and then treated with 10 % HCl solution to enable fluoride dissolution. We then concentrated pollen by physical enrichment procedures, using ZnCl 2 separation and ultrasound sieving over a 10 µm filter. Samples were stored in glycerin. Identifications were based on atlas of pollen and spores (Wang, 1995;Song, 1999), as well as modern reference slides from the collection of the Laboratory of Sporopollen Analysis of the Geography Department of Lanzhou University. A palynological diagram was plotted using Tilia v2.0.b.4 (Grimm, 1993) and pollen-assemblage zones were constructed using stratigraphically constrained cluster analysis (CONISS) (Grimm, 1987).

Results
Only 126 of the 200 samples contained enough palynomorphs to provide reliable data; the remaining 74 possessed fewer than 300 identifiable grains and have not been included in the analysis. Most of the latter samples had been preserved under oxidizing conditions or had high carbonate content. Approximately 80 different palynomorphs were identified at family or genus level. Percentages were expressed on the total number of recognized taxa. Tree pollen consists mainly of Pinus, Cupressaceae and Ulmus, along with Quercus and Betula. Additionally, a number of subtropical plants pollen, such as Liquidambar, Pterocarya and Carya (which are no longer found in this area today), appear often in low abundance. Herbaceous pollen is mainly from Artemisia, Chenopodioideae, Poaceae and Asteraceae. Pollen from extremely drought-tolerant plants, such as Ephedra and Nitraria, only appear sporadically in single samples. In addition, the section also contains fern spores and Pediastrum colonies. A selection of the more important taxa is given in Fig. 3. CONISS (Grimm, 1987) yields three distinct zones, described from the bottom up as follows.

Vegetation and climate reconstruction
The sedimentary facies of the Yaodian Section indicate four successive depositional stages: fluvial channel, floodplain, shallow lake, and distal floodplain to palustrine. Transitionals can be dated to 10.4, 9.23 and 7.43 Ma  ( Fig. 2). Our palynological record shows stepwise changes at 10.1 and 7.4 Ma, lagging slightly behind those evinced by the sedimentary facies. Another distinctive feature of the palynological record is that the green lacustrine deposits of fine grain size exhibit dense palynomorph concentrations, with higher tree pollen percentages. In contrast, the reddish floodplain deposits with coarse grain sizes possess sparse palynomorph concentrations, with higher herbaceous pollen percentages (Fig. 3). However, in the same pollen zones, we find that the palynomorph concentration clearly changes between different sedimentary facies, but that percentage fluctuations are minor. Between different pollen zones, the palynomorph percentages change strongly within the same sedimentary facies. We can therefore conclude that the changes in the palynological record are caused by changes in regional vegetation, rather than different preservation conditions. The paleoecological information inferred from the percentage change of pollen record can thus be considered reliable. According to modern surface pollen studies, Pinus is often overrepresented in pollen records because of its abundant pollen production and the ease with which this pollen is transported over long distances. As a general rule, it can be assumed that there is/was no proximate pine forest if less than 25 to 30 % of Pinus pollen occurs in samples (Li and Yao, 1990). Higher percentages of Cupressaceae and Taxodiaceae coexistent with temperate tree, shrub and herbaceous pollen may reflect a warmer, wetter and more humid climate (Song, 1978). Nowadays, Ulmus is commonly distributed in the subhumid temperate and warm-temperate mountain foothills of northern China, but percentages of its pollen collected from the Chinese Loess Plateau surface soils never exceed 1 %, even under broadleaved forests containing elm (Liu et al., 1999). In general, when their abundance exceeds 3-5 % of the arboreal pollen total, birch and oak can be considered to be/have been present in woodland (Liu et al., 1999). Salix produces very little pollen, and most of this pollen falls near the tree itself (Li et al., 2000). Modern Artemisia and Chenopodioideae are extensively distributed throughout the arid and semi-arid regions of China. Chenopodioideae are more drought-resistant than Artemisia. Higher percentages of Artemisia pollen may reflect a semiarid grassland environment, while higher percentages of Chenopodioideae pollen may reflect an arid desert environment. Surface pollen analysis shows that Artemisia and Chenopodioideae are greatly overrepresented in the pollen rain. Only when Chenopodioideae and Artemisia pollen abundance exceeds 30 % of the total should their presence be considered primarily local (Herzschuh et al., 2003;Ma et al., 2008). Poaceae pollen abundance is sparse, usually only 3-6 %, even when it represents the dominant modern species (Tong et al., 1995).
Our record therefore indicates that, during the period when the Yaodian Fm was being deposited, the study area was covered by temperate forests and a warm and humid cli-   (Enkelmann et al., 2006;Fang et al., 2003Fang et al., , 2005Lease et al., 2007;Li et al., 2014;Molnar et al., 2010;Wang et al., 2006;Zheng et al., 2006Zheng et al., , 2010   also support the theory that much denser vegetation capable of supporting large mammals such as rhinoceroses developed during this interval. Moreover, we know that the northern Tianshui Basin was dominated by temperate and warm-temperate deciduous broadleaf forest (Hui et al., 2011). Our result is also consistent with research into the climatic evolution of the Qaidam Basin, which found that the presence of δ 18 O values characteristic of large mammals indicated a warmer, wetter, and perhaps lower-altitude Qaidam Basin . The early late Miocene mammal fauna discovered in the Qaidam Basin also reflects a wooded environment, in which many streams with aquatic plants such as Trapa and Typha developed . From 10.1 to 7.4 Ma, the study area was dominated by a warm-temperate open forest environment and a less humid climate, relative to the previous interval. Sedimentary facies become characteristic of shallow lake deposits . Mammal fauna identified in the eastern Qaidam Basin also indicates that a mixed habitat of open and wooded environments, with abundant freshwater streams, was predominant at that time . In particular, herbaceous plants also increased their presence in the Tianshui Basin after ∼ 8.6 Ma, as confirmed by mammalian fossil records. In the northern Tianshui Basin at ∼ 9.5 Ma, there is evidence of a sizeable rhinoceros population, which would have required a relatively moist woodland environment to sustain itself. However, the typical Hipparion fauna at ∼ 8.0 Ma probably represents a relatively temperate climate with more mixed vegetation, i.e., an open forest environment rather than a vast, open landscape. Large mammals would still have been able to survive in such an environment ). An open temperate forest-steppe environment developed in the study region, indicating significant aridification after ∼ 7.4 Ma. Grassland, composed principally of Poaceae, Artemisia and Chenopodioideae, developed in most of the basin, while shrinking areas of open forest, dominated by Cupressaceae, Ulmus and Quercus, existed in the surrounding mountains. Salix continued to grow in relatively humid environments such as riverbanks. Distal floodplain to palustrine deposits now characterized the study area . A sudden increase in magnetic susceptibility after ∼ 7.4 Ma may indicate an arid environment (Zhang, 2013) (Fig. 4b).
In the northern part of the Tianshui Basin, drought-tolerant Artemisia predominated after 7.4 Ma, further confirming the presence of a drier climate (Hui et al., 2011) (Fig. 4c). Additionally, the growing presence of grazer mammalian species at the end of the Miocene in the Tianshui Basin suggests that the local environment was principally occupied by grassland, with some woodland, and even some deserts (L. P.   (Fig. 4d). Furthermore, the gradual increase in eolian sediments after 7.4 Ma in the Linxia Basin would indicate a period of intense desertification in central China (Fan et al., 2006) (Fig. 4e). Biomarker evidence from the Linxia Basin also indicates a distinct change in climate toward arid-cold conditions at ∼ 8 Ma (Y. L. . The isotopic compositions of herbivorous fossil teeth and paleosols from the Linxia Basin (Wang and Deng, 2005) and southwestern China (Biasatti et al., 2012) also indicate a shift to a drier, or seasonally drier, local climate. In the Qaidam Basin, Hipparion teilhardi fossils are characterized by slenderer distal limbs and are dated to the end of the Miocene, implying an adaptation by this animal to the open steppe environment (Deng and Wang, 2004). Marine sediments also indicate that the climate changed at this time. For example, local seawater δ 18 O reconstructions from ODP Site 1146 in the northern South China Sea suggest that the climate of east and south Asia shifted toward more arid conditions after ∼ 7.5 Ma (Steinke et al., 2010) (Fig. 4f).

More arid condition at the end of the Miocene and possible causes
Based on the late Neogene Chinese mammalian fossils data, Zhang (2006) suggested that mammal communities in northern China were rather stable and uniform from ∼ 13 Ma to the end of the Miocene (∼ 7-8 Ma), and that differentiation between the humid fauna communities prevalent in eastern China and the dry fauna communities identified in western China occurred after the end of the Miocene. The diversity in Bovidae fossils also increases significantly toward the end of the Miocene, with some genera appearing in southwestern China (Chen and Zhang, 2009), indicating an expansion of grasslands and aridification. Using macro-and microfloral quantitative recovery techniques to reconstruct the climate in northern China at the time, Y.-S. C.  proposed that the west-east temperature and precipitation gradient pattern did not develop in northern China until the end of the Miocene. This corroborates the quantitative results gained from using mammalian fossils as a proxy for paleoprecipitation (Liu et al., 2009). A semi-quantitative reconstruction of Chinese Neogene vegetation also indicated that the aridification of western, central and northern China occurred during the Miocene-Pliocene transition (Jacques et al., 2013). Indeed, in order to adapt to the arid climate of northern China during the end of the Miocene, some plants and arthropods also evolved more aridtolerant species, such as Frutescentes (Fabaceae) (Zhang and Fritsch, 2010), Ephedra (Ephedraceae) (Qin et al., 2013) and Mesobuthus (Buthidae) . This marked aridification has been well documented in other parts of Asia. For example, dramatic changes in the carbon isotopic ratio of leaf waxes at ODP Site 722 indicate an increasing aridity at the end of the Miocene in continental source regions, including Pakistan, Iran, Afghanistan, and the Arabian Peninsula (Huang et al., 2007) (Fig. 4g). The isotopic compositions of herbivorous fossil teeth and paleosol carbonates also suggest that the climate became drier over the Indian Subcontinent, China, and central Asia toward the end of the Miocene (Badgley et al., 2008;Barry et al., 2002;Biasatti et al., 2012;Cerling et al., 1997;Quade et al., 1989;Wang and Deng, 2005;Zhang et al., 2009). The evidential synchronicity of these climatic events in Asia strongly suggests that the aridification of the Asian interior began at the end of the Miocene (∼ 7-8 Ma). The onset of such a marked aridification is further corroborated by the presence of red clay across much of the Chinese Loess Plateau . Precipitation in arid northwestern China is primarily caused by the Asian summer monsoon, whereas the Asian winter monsoon promotes a cold and dry climate. Besides the monsoon source, the westerlies also bring precipitation into China. During the Neogene, Eurasia was influenced by global cooling, land-sea redistribution and regional tectonic uplift (Lease et al., 2007;Li et al., 2014;Guo et al., 2008;Miao et al., 2013Miao et al., , 2016Molnar et al., 2010;Mudelsee et al., 2014;Zachos et al., 2001;Zhang et al., 2007), and these three factors are considered as the major drivers for the formation and evolution of the Asian monsoon and inland arid climate.
During the late Neogene, the most significant global cooling event occurred at ∼ 14 Ma (Mudelsee et al., 2014;Zachos et al., 2001), followed by a longer-term but minor cooling trend (4-10 Ma; Mudelsee et al., 2014) (Fig. 4h). Although the global cooling should somehow lead to net aridification on the planet, cooling and aridification trends do not seem to run parallel (van Dam, 2006). The complexity of the atmospheric and oceanic circulation systems ensures that general cooling may result in a precipitation decrease in some regions and an increase in others (van Dam, 2006). However, integrated studies have showed that the global cooling during the Neogene had significant influences on driving the Asian monsoon and inland arid climate (e.g., Lu et al., 2010;Lu and Guo, 2014;Tang and Ding, 2013), especially since the late Miocene (Lu and Guo, 2014). The possible mechanism lies in two aspects. Firstly, it is clear that the global cooling has strengthened the Siberia High, which dominates winter monsoon circulation and aridity in eastern Asia (Lu and Guo, 2014). This would result in enhanced and more frequent cold surges in the mid-latitudes of the Northern Hemisphere. Secondly, the global cooling caused the weakening of hydrological cycle, expanding of ice sheets, lowering of sea level and increasing of continental surface (Lu and Guo, 2014;Tang and Ding, 2013). This would reduce the moisture mass transported into the continental interior (Tang and Ding, 2013). Therefore, we speculate that the global cooling could intensify aridity of the Asian interior.
Besides the above focusing on the climate effects of the global cooling, model simulations have paid special attention to the climatic effects of the land-sea redistribution. For example, model simulations suggest that the westward retreat of the Paratethys from central Asia has contributed significantly to Asian climates (e.g., Guo et al., 2008;Ramstein et al., 1997;Zhang et al., 2007). However, a large amount of geological evidence suggests that the vast majority, or even all, of the Paratethys regression from the Tarim Basin (northwestern China) occurred at the Oligocene (e.g., Bershaw et al., 2012;Bosboom et al., 2014). At the same time, numerical simulation also indicates that the spreading of the South China Sea may enhance the south-north contrast of humidity in China (Guo et al., 2008), bringing more precipitation into Asia. Nevertheless, many studies indicate that western and northern China became drier during the Neogene (e.g., Guo et al., 2008;Tang and Ding, 2013;Sun and Wang, 2005). Therefore, although the land-sea redistribution had a significant impact on the major climate reorganization in Asia during the late Oligocene/early Miocene (Guo et al., 2008;Zhang et al., 2007), it should have a limited effect on the formation and development of the Asian inland arid climate during the late Miocene.
Model simulations have also paid attention to the climate effects of the TP uplift. The scenarios of whole-plateau up-lift (e.g., Kutzbach et al., 1993), phased uplift (e.g., An et al., 2001;Kitoh, 2004;Liu and Yin, 2002) and subregional uplift (e.g., Kuang, 2010, 2013;Chen et al., 2014;Tang et al., 2011Tang et al., , 2013Wu et al., 2012), with increasing complexity, are usually designed for discovering the cause-effect relations between the plateau uplift and paleoclimate change (Liu and Yin, 2011). The different models conclude that the uplift of the TP played an essential role in affecting the atmospheric circulation and forming the monsoon and arid climate when the whole/subregional plateau exceeds a critical height Kuang, 2010, 2013;Chen et al., 2014;Kutzbach et al., 1993;Liu and Yin, 2002;Tang et al., 2011Tang et al., , 2013Wu et al., 2012). However, because of the different model setups and boundary conditions, there still exist many uncertainties in the different forms of the plateau uplift forcing and regional climatic responses (Liu and Yin, 2011). The geological/proxy research can provide the constraints for the model boundary conditions, whereas numerical simulation can test the geological/proxy result. Therefore, it is useful to compare the geological/proxy results and the numerical simulations (Micheels et al., 2007. Many geological studies have suggested that the TP experienced rapid uplift during the interval ∼ 8-10 Ma (e.g., Enkelmann et al., 2006;Fang et al., 2003Fang et al., , 2005Lease et al., 2007;Li et al., 2014;Molnar et al., 2010;Wang et al., 2006;Zheng et al., 2006Zheng et al., , 2010 (Fig. 4i), but the timing and degree of the uplift are still debated. The late Miocene uplift would have achieved an altitude sufficient to block the penetration of moisture from the source region into western China (Dettman et al., 2001(Dettman et al., , 2003. There is also increasing proxy evidence that the Asian summer monsoon weakened after ∼ 10 Ma (e.g., Clift et al., 2008;Wan et al., 2010), while the Asian winter monsoon strengthened, particularly toward the end of the Miocene (e.g., An et al., 2001;Clift et al., 2008;Jacques et al., 2013;Jia et al., 2003;Sun and Wang, 2005), implying the involvement of intensified Asian inland aridification. It is consistent with the most model simulations that aridity of the Asian interior will be intensified along with the uplift of the TP. However, it should be noted that there is no doubt regarding the effects of the global cooling on the general trend toward a dry climate in the Asian interior.

Conclusions
The late Cenozoic basins, located at the northeast TP, document the environmental changes associated with tectonic uplift and global cooling. We investigate a late Miocene pollen record from the Tianshui Basin. Our results indicate that a temperate forest with a rather humid climate regime (11.4-10.1 Ma) gave way to a temperate open forest environment with a less humid climate (10.1-7.4 Ma); this was in turn replaced by an open temperate forest-steppe landscape, accompanied by a relatively arid climate (7.4-6.4 Ma). The vegetation succession demonstrates that the aridification of the Asian interior occurred after ∼ 7-8 Ma, as corroborated by other studies of Asia. Our findings support the idea that the long-term global cooling and the TP uplift caused the late Miocene aridification of the Asian interior.

Data availability
Data used in this study can be found in the Supplement.
The Supplement related to this article is available online at doi:10.5194/cp-12-1473-2016-supplement.